LINGUIST List 13.1642

Mon Jun 10 2002

Sum: Consonant Harmony

Editor for this issue: Marie Klopfenstein <>


  1. Dan Everett, Consonant Harmony

Message 1: Consonant Harmony

Date: Sat, 8 Jun 2002 11:35:44 -0300
From: Dan Everett <>
Subject: Consonant Harmony

A couple of days ago I posted a question on Consonant Harmony (CH) in
adult phonologies. I asked the question this way:

Does anyone know of clear cases of Consonant Harmony occurring in adult
Some researchers have observed CH in L1 acquisition data, but others
have claimed that it cannot appear in adult speech. I reported on an
apparent case in: 1985 'Syllable Weight, Sloppy Phonemes, and Channels
in Pirah´┐Ż Discourse,' In: Mary Niepokuj (eds.) Proceedings of the
Berkeley Linguistics Society 11, pp 408-416. 
In child language acquisition cases might look like: [mom] for German
/Baum/ or [guk] for /buk/ 'book', etc. In Piraha it looks like ?apapai
'head' ~ ?a?a?ai 'head' (?=glottal); Kohoi 'name' ~ Kokoi, etc."
I received a few responses, all useful, and I thank the respondents for
taking the time to help out. Most respondents noted that my question was
not all that clear, because CH has many different forms. As they deduced
correctly from my examples, however, I was after primary place harmony.
Here is what I learned. I hope it is useful to some LL readers as well:

>From Vern M. Lindblad: "... some Turkic lgs. have an alternation pattern
that I would consider to be a form of consonant harmony. Along with the well-known patterns of
backness harmony in vowels that are found in most Turkic lgs., some also have
velar consonants in suffixes in front harmonic environments alternating with
the corresponding uvular consonants in back harmonic environments. The
specific example I know best is Uyghur, but I seem to remember that others show 
similar patterns. My guess would be that any Turkic lg. with both
velars and uvulars would probably display this pattern."

>From Adam Werle: "... it sounds from your inquiry on the Linguist List
like you're mainly interested in adult consonant harmony of major place (i.e. labial, 
coronal, dorsal). But if you're also interested in secondary place C 
harmony (e.g. lip rounding, palatality, etc.), you may be interested 
in Gunnar Hansson's ( dissertation, which is a 
survey of C harmony systems across languages. One of his findings, I 
think, is that major place C harmony is unattested in adult language."

The reference to Hansson's thesis was quite useful. Before I could
pursue it, Hansson himself wrote, summarizing his research. His research
seems to be the most recent and most exhaustive work on the subject and
his thesis is soon to be published by CSLI, as I understand it. Here is
what Gunnar wrote:

" From the discussion in your query, it appears that by "consonant
harmony" you are referring specifically to *place* harmony--long-distance 
agreement in (primary) place of articulation. This is what is so 
prevalent in child language (although other types do occur as well) 
and at the same time seems to be unattested in adult language.

I just wanted to emphasize that other types of CH do occur in adult 
language, involving a wide range of properties: laryngeal features 
(voicing, glottalization, aspiration, or some combination of these), 
nasality (e.g., in many Bantu lgs, where nasals and voiced stops 
interact without any concomitant nasalization of intervening 
segments), laterality/rhoticity (interaction of laterals and 
rhotics), stricture (interaction of fricatives and stops, or 
fricatives and affricates), secondary articulation 
(pharyngealization, velarization, labialization, etc., again without 
affecting intervening segments), and so forth. At the end of the 
scale that's closest to the primary-place harmony of child language, 
you also find various types of "minor-place" CH. The most obvious and 
most common type, of course, is coronal harmony (typically sibilant 
harmony, but many other varieties exist), but there are also several 
cases of "dorsal harmony", where uvulars and velars are involved in 
long-distance assimilation (as before, without intervening segments 
being affected).

You may want to have a look at my 2001 UC Berkeley dissertation, 
which discusses at great length the typology of attested CH systems 
and cites a great amount of data culled from the descriptive 

Returning to (major-)place consonant harmony, it does seem to be 
unattested--at least as a regular phonological process--in adult 
language, as far as I know. Nevertheless, I point out in the 
dissertation that a great number of languages have MSCs that are 
*usually* stated as laryngeal dissimilation, but which could equally 
well be expressed as a matter of place assimilation/harmony 
(parasitic on laryngeal features). For example, many languages allow 
homorganic glottalic-glottalic combinations (/k'Vk'V/, /t'Vt'V/, 
etc.) as well as heterorganic glottalic-pulmonic combinations 
(/k'VtV/, /t'VpV/ etc.), but crucially exclude heterorganic 
glottalic-glottalic combinations. (A number of these are discussed in 
Margaret MacEachern's 1997 UCLA dissertation.) The usual way of 
phrasing this generalization--e.g., in Yip's (1989) discussion of 
Yucatec--has been that "if two consonants differ in place, they must 
also differ in laryngeal features" (or vice versa). But it is equally 
possible to describe it as a requirement that "if two consonants 
agree in laryngeal features, they must also agree in place", which is 
then by definition a case of (static) consonant place harmony. From a 
synchronic-descriptive standpoint, the two are equivalent, but a 
relevant question to ask is, of course, what really happened 
historically that led to this synchronic pattern (assimilation or 

Others who wrote were Lameen Souag, who said: "...emphatic spread in
Arabic dialects (eg Algerian r.aas > r.aas., Egyptian t.alab > seems
like a pretty clear case of consonant harmony to me... I don't know any 
references to recommend offhand, but any discussion of Arabic phonology
or Arabic dialectal phonology should incorporate it."

Mikael Parkvall mentioned cases of CH in Rajasthani and some North
American languages: "...* Some varieties of Rajasthani (and apparently
also some other Indo-Aryan lgs) has consonantal harmony (Masica 1991:130).
* Some North American American languages too, are said to have it, in 
particular Athabaskan ones (Mithun 1999:26ff, 361). Mithun, Marianne
(1999): The languages of Native North America. Cambridge: Cambridge
University Press. Masica, Colin (1991): The Indo-Aryan languages.
Cambridge: Cambridge 
University Press."

Stephan Schmid reported that "... Laver (1994: 388-9) claims that there
be CH in the Niger-Kongo language Estako, establishing three consonant
harmony sets, i.e. tense, lax, and neutral."

John Koontz offered some general comments on separating CH from Vowel
Harmony by focussing on consonantal feature spread, rather than features
which could be attributed to Vowel Harmony or word-level phenomena
incidentally affecting consonants. 

Again, I want to thank those that responded. There is some interesting
new work on CH in child-language acquisition in progress by Marilyn
Vihman and Phil Carr as well, reported at the recent 10th Annual
Manchester Phonology Meeting.

Dan Everett

P.S. GO BRAZIL - win that Cup!
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue